« first day (3090 days earlier)      last day (689 days later) » 

12:30 AM
tex live 2019 soon ^_^. Been waiting to install it on my new computer until 2019 launches
I made /usr/local/texlive its own subvolume so it won't show up in my snapshots though :P
5 hours later…
5:44 AM
Can someone please test this with the latex+dvips+ps2pdf for me?
\tikz \shade[shading=Mandelbrot set] (0,0) rectangle (2,2);
I suddenly get GhostScript errors :( after an upgrade (Debian testing). But I can't figure out what is to blame. I don't think it's GhostScript.
5:58 AM
@DavidPurton I get also Ghostscript errors on my TeXLive 2018 distribution on a Mac.
6:09 AM
@marmot At least it's not just me! But this was working. Nothing from TeXLive was upgraded. GhostScript went from 9.26 to 9.27, but downgrading does not help. There doesn't seem to be any other major updates done. This is also the only Type 4 shading from the tikz shadings library that fails. Odd.
@DavidPurton Maybe this happens only to TikZlings? ;-)
@HenriMenke, do you have any clue about this Mandelbrot set problem? Can you reproduce it?
6:36 AM
Interesting. It works (sort of—the output looks wrong, but doesn't give errors) up to 9 iterations. But adding the tenth causes GhostScript errors.
Maybe it never worked, and I just thought I tested it :(.
6:56 AM
@marmot I have some edit to suggest to your post tex.stackexchange.com/questions/423019/… about errors in the pgfmanual. Feel free to accept or reject it (as I have low reputation, the edit will be send in the queue for peer review). Can I send it?
@marmot I have send.
1 hour later…
8:24 AM
> Alexandra Kehayoglou makes carpets that evoke the topography of her native Argentina: grasslands, waterways, and glaciers. Her family opened a conventional carpet company there in 1956, and she discovered she could use scraps from their work to create “tactile canvases.” Each piece is composed by hand from discarded or surplus wool on a vertical frame, using a tufting gun and carpet scissors.
2 hours later…
10:14 AM
Recently, I've come across many "reference-like" questions (common problem, high number of votes and views) whose answers unfortunately suffer from clutter: a) inconsistent formatting/indenting, b) too much focus on the OP's original code.

E.g. take https://tex.stackexchange.com/questions/278727/split-subfigures-over-multiple-pages/278748#278748. The actual answer to the problem could be stated in ~10 lines of LaTeX code.
Or take tex.stackexchange.com/q/188650/38074: I spent a few minutes trying to get the top answer to work on my code and the problem was that I could not easily distinguish between [code from OP in answer] and [required code for solution]. Again, the actual answer could be stated in way less code, e.g. see my answer there.
StackExchange draws a considerable amount of its value from finding such Q & A via search engines. I feel like we should also strive to optimize Q & A posts for that. What does the community think of this?
(Also, one of recent edits intended to make it easier for users coming from search engines was rejected. That's why I'd like to hear the community's opinion on that before :))
10:31 AM
@ComFreek One for a meta post? Get a wider range of views?
@ComFreek There's a balance in terms of making sure the OP gets the information they actually need, I guess
@ComFreek lots of questions could be answered in one line "read the manual" but if people feel the need to make a specific example and ask a specific question, it is best to give them an answer using that example.
@ComFreek why would that edit make it easier for search engines? It mostly seemed to consist of changing the position of \label (which is actually an improvement, but not direct;y related to the question)
@DavidPurton I think this is a stack overflow, but I have to dive deeper to tell.
@DavidPurton Are the other bugs fixed to your satisfaction?
@ComFreek the posted answer is only 8 or 9 lines of code for a complete commutative diagram I don't think you could call that in any way excessive even if it could in theory be compressed more.
@DavidPurton The dvipdfm problem with shadings not being able to be redeclared is annoying. It might have actually been there for quite some time, because I don't think anyone is using dvipdfm. Maybe I'll just deprecate it.
@HenriMenke does it also apply to xdvipdfmx (in which case that is all xetex users)
10:40 AM
@DavidCarlisle Usually it does ...
@JosephWright i know:-)
@HenriMenke (u)pTeX: anyone Japanese, more or less
@DavidCarlisle I don't think so.
@JosephWright dvipdfm and dvipdfmx are separate drivers in PGF. Does (u)pTeX really use dvipdfm rather than dvipdfmx?
@HenriMenke Oh, sorry: I thought you meant dvipdfmx
@JosephWright It easy to get confused with L3 having an x in all the names. ;)
10:45 AM
Is dvipdfm now also xdvipdfmx internally? I got a warning from xdvipdfmx when trying the example.
@HenriMenke More that as you say more or less no-one is using dvipdfm
@UlrikeFischer dvipdfmx and xdvipdfmx are being maintained together, so there are only a few differences
@JosephWright That is what I thought.
@JosephWright I know, but I asked about dvipdfm without the x at the end.
@HenriMenke We don't provide a driver for example in the core graphics distro
G:\Z-Test>dvipdfm test-utf8.dvi
test-utf8.dvi -> test-utf8.pdf
xdvipdfmx:warning: Object @pgfshademyshadingA already defined.
2318 bytes written
10:48 AM
@HenriMenke graphics package does:
just aliasing the names.
@DavidCarlisle Thanks for the info. I will try to merge them.
G:\Z-Test>dvipdfm --help

Usage: xdvipdfmx [OPTION]... [DVIFILE[.dvi|.xdv]]
@HenriMenke it looks as if dvipdfm is actually xdvipdfmx (at least in tl19)
@UlrikeFischer That might be the origin of many of those errors.
lrwxrwxrwx 1 davidc NAGNTD+Group(513) 13 Oct  3  2018 /usr/local/texlive/2019/bin/x86_64-cygwin/dvipdfm -> xdvipdfmx.exe
There is an option ` --dvipdfm Enable DVIPDFM emulation mode`.
10:51 AM
@HenriMenke @UlrikeFischer ^^
@DavidCarlisle Sure, giving an answer tailored to OP's specific code is good! But that only helps OP in most cases. All other users would benefit more from more minimal examples, I suppose.
I corrected the indentation in the edit. (The \label thing I did just do since I already was at editing.) At least for me that makes it easier to read and separate OP-specific code from solution-specific code.
E.g. the answer says "\arrow[thick]{l}[name=D]{\pi^\star}", then I thought "alright, let's adapt it to my code as follows:" \arrow[l, name=D], but that didn't work! You have to use \arrow[]{r}[name=D]{}.
That was totally unclear to me. Sure, reading the manual would have helped, but also required a lot more time.
@UlrikeFischer I have a plan on the link business: i think we can avoid needing to grab arguments entirely
@ComFreek Like I said, I think this probably needs a meta discussion
11:16 AM
Alright, I'll create a meta post later today.
11:30 AM
@ComFreek I'd agree with the general principle of what you are saying but don't think the examples you picked are good, if the answer contains (as they do sometimes) hundreds of lines of irrelevant preamble copied from the OPs example in the question then yes it obscures it for others, but if the code in the answer is a complete CD in ten or so lines of tikz then it may or may not be a good answer but I don't think you can complain about it not being minimal.
@ComFreek you changed the argument structure completely and it didn't work, bad luck:-) I'd put that down to user error not a problem with the answer as posted.
11:57 AM
@HenriMenke yes, I think the other bugs are good. I rebased the CMYK patches and everything seems to work (except the Mandelbrot set for dvips). But the more I think about this, I don't think I tested it, because there were other bugs with the dvips driver.
@DavidCarlisle no, the xdvipdfmx driver is fine.
@HenriMenke deprecating is OK. I was just testing everything that is supposed to be supported and noticed this.
@DavidPurton as noted above dvipdfm is a symbolic link to xdvipdfmx (but the name forces some compatibility defaults) so probably pgf should be using the same driver macros for dvipdfm and dvipdfmx
12:12 PM
@UlrikeFischer, @DavidCarlisle Comments on latest checks for dvips links welcome
@JosephWright I would say it is a pdf concept. accsupp uses \special{pdf:content EMC} (dvipdfm), and \special{ps:[/EMC pdfmark} (dvips) and \pdfliteral page{EMC} with pdftex. (The slash in the dvips case is interesting, do one have to add it to operators with dvips? I wonder how to handle this ...).
@UlrikeFischer dvips is using pdfmark, which means we have to have operators
@UlrikeFischer Don't we want to abstract those as some kind of EMC function then?
12:29 PM
@UlrikeFischer Yes, this is the pdfmark syntax for marked content. See help.adobe.com/en_US/acrobat/acrobat_dc_sdk/2015/HTMLHelp/…
@JosephWright I have also BDC (which has two argument) and BMC (1 argument) operators, in ocgx2 I found stuff like \pbs_literal:nn{page}{q~7~Tr}. pdfbase has also commands like \pbs_pdfemc:.
@AlexG You beat me to it!
@UlrikeFischer I really think we want a separate function for 'generic page' stuff if required, but otherwise prefer something like \driver_pdf_mark_end or similar for EMC
@JosephWright Some special functions for the three operators would be fine as they are used a lot (but I would use bmc, bdc and emc in the name - easier to connect). But beside this: the \pbs_pdfbdc:nn from @AlexG looks a bit more "higher up", it also writes stuff to the page/object resources. Something that neither I nor accsupp has.
@UlrikeFischer / is postscript quote so /EMC is the name EMC rather than trying to execute whatever EMC is bound to (not needed in the pdf cases as there there are custom syntax that expect a name at that point)
12:37 PM
@DavidCarlisle Ah. At the end of the year I will probably know postscript too ;-)
@UlrikeFischer at the end of last century I knew something, now less so...
12:53 PM
@UlrikeFischer Yes BDC/EMC are also relevant for marking optional content (OCGs). The BDC/EMC pdfmarks have been implemented first in in gs-9.14 after I provided a corresponding patch to create PDF Layers with dvips+ps2pdf: git.ghostscript.com/…
@AlexG I saw that, the main question is what sort of commands/abstraction are useful in expl3. Imho \driver_pdf_literal_bmc:n, \driver_pdf_literal_bdc:n and \driver_pdf_literal_emc: and perhaps a generic \driver_pdf_literal_page:n would be ok (@JosephWright).
1:12 PM
@DavidCarlisle So I see. And it does work fine just to use the dvipdfmx driver for dvipdfm.
@UlrikeFischer I'd say the implementation of \pbs_pdfbdc:nn and \pbs_pdfemc: in pdfbase.sty is quite generic for all drivers. Note that BDC has two mandatory aruments (acc. to the pdfmark reference): a tag and a dictionary. The BDC pdfmark automatically creates a PDF object from the dictionary, maps the obj number to a /name in the resources dictionary of the content stream and finally inserts /tag /name BDC in the content stream (i. e. page or XObject).
The mapping is necessary, as objects cannot be referred to by their object number in content streams. For pdftex and dvipdfmx drivers this has to be done manually. See pdfbase.sty.
@DavidPurton tell @HenriMenke :-)
@AlexG I know that BDC has two arguments, but I don't need object, for tagging there is simply a number in the dictionary and then the parent tree is used to find it.
@UlrikeFischer Oh sorry for the noise!
@DavidCarlisle I commented on the bug report
1:20 PM
@DavidPurton thanks (ug report sounded better:-)
@AlexG no it is interesting. After all we are trying to build pdfresources useful for all, so other use cases should be looked at too (e.g. also accsupp etc).
@DavidCarlisle I have a Macook Air and key doesn't work properly…
@UlrikeFischer Did you try \pbs_pdfbdc:nn ... \pbs_pdfemc: and look whether it works in your case?
@DavidPurton I have a dell laptop and it always writes the as teh, I have no idea why
@DavidCarlisle try hunting and pecking more slowly
@DavidCarlisle the b key works if I hit it hard enough :)
@HenriMenke Ugh. Those PostScript functions are a pain. When I worked out the original code for pgf-cmykshadings I had to painstakingly work things out using the gs interpreter and print the stack all the time. sigh
1:36 PM
@UlrikeFischer Could you please post here a short example that demonstrates how the opening and closing tags would have to look like in the PDF content stream? I guess something similar to /some_tag_name << a dictionary >> BDC bla bla blub EMC.
@AlexG A typical BDC looks like this: /Lbl<</MCID 11>> BDC. The numbers are per page and I have to keep track of them, as they are referenced in the parenttree. The dictionary can contain more stuff (e.g. the /Alt or /Actualtext). The end is simply EMC. It could be (if a complete xobject is marked) that I at some time will need to create objects and objects reference too.
@quark67 Thanks! (Yes, when the pgfmanual changes, then these references are no longer accurate but it is not clear to me whether one should edit all answers to reflect such changes.)
1:59 PM
@UlrikeFischer Using pdfbase it could look as follows:



\pbsDictObj{/MCID 11}%
\pbsPdfBDC{/Lbl}{\pbsLastObj}Bla, bla, blub\pbsPdfEMC

2:09 PM
Or even shorter:



\pbsPdfBDC{/Lbl}{<</MCID 11>>}Bla, bla, blub\pbsPdfEMC

2:22 PM
@AlexG In the long run I should certainly move the dictionary out of the page stream and use a named object instead. But 1. I lied a bit, I'm not always using \pdfliteral, but in lua it is actually done by injecting nodes, so I need to check how to do something similar there too (shouldn't be too difficult). And 2. I want to base the code on l3kernel code (and so push the development there), so your code (at least the basic stuff) should move into l3drivers ;-)
@UlrikeFischer Ah, nice! In luaTeX, I think , there is a direct replacement for \pdfliteral, without the need for using lua code?
@AlexG yes, but I need the lua to inject the literals at shipout time (I wander through the output box and check attributes). This avoids that one ends with unbalanced (not closed) BDC operators on a page.
@UlrikeFischer Ah, interesting. With the ocgx2 package I put a lot of effort to avoid the same for OCGs (PDF Layers) broken across page breaks. But I did this on the expl3(-TeX)-level.
By maintaining stacks of open BDCs.
@AlexG what are you doing to close the BDC? Do you have the case that you have to handle a complete paragraph with a page break in it?
2:40 PM
For every open BDC I put some marker on a stack (L3 seq) (Yes, PDF layers may be nested.). In the page output routine I check whether the stack is empty or not. If it is not empty I close all of them. Then, on the next page, I reopen them in the same order.
(Closing with \pbs_pdfemc:)
@UlrikeFischer see file ocgx2.sty : function \ocgxii_beginocg:nnnn . Somewhere at the end of this function, there is \ocgbase_open_stack_push:n{...}
@AlexG my problem is that I can have documents like the following, and I need to put on page 1 an EMC after the "consectetuer" of the main text, and the "congue" in the footnote - and I didn't add floats yet.
\driver_pdf_compresslevel:n {0}
\textheight 5cm
\pdfliteral page {/P <<MCID 0>> BDC}
\footnote{\pdfliteral page {/P <<MCID 1>> BDC}\lipsum[1]\pdfliteral page{EMC}}
\pdfliteral page{EMC}
@AlexG so it is not about finding out if there is an open BDC but to find the right place to close it (and to inject there something).
@UlrikeFischer vv
Don't use footnotes in your books, Don.
\author JILL ^{KNUTH} (1962)
@UlrikeFischer Oh, sounds complicated, and dangerous, in particular if a long footnote text is wrapped around a page break. So it is better to not use footnotes, as @DavidCarlisle suggests. ;-)
@UlrikeFischer It is mandatory to close ANY BDC on the SAME page with a matching EMC. Otherwise, related BDC/EMC end up in different content streams.
2:58 PM
@AlexG well footnotes are only a rather dramatic example, floats, headers, longtables and other stuff moving around or changing the output routine are problematic too.
Therefore you must maintain a stack of open BDCs to close them at the page end and reopen at the beginning of the next page.
@AlexG from the tagpdf docu ^^^
@AlexG did you see the document posted to a question on the main site the other day? One multi-page footnote and on each page just two or three lines of normal text with the rest of the page being footnote...
@UlrikeFischer A float shouldn't be a problem, I guess, because it lives on a single page.
@DavidCarlisle Must have looked very funny.
@AlexG no the float itself is easy, but if there is a float it gets more difficult to find the place to restart the marking of the normal text. If there were only text, I could simply use atbegshi and add something at the begin and end of the shipout box.
3:03 PM
@AlexG trying to honour the layout of an ancient manuscript so unfortunately "use a sensible layout" not an acceptable answer.
Especially. lawyers are fond of this kind of texts.;)
@AlexG yes they don't want the text at a size their clients can read.
@DavidCarlisle Lol
@DavidCarlisle it is always nice when you can blame someone from the 17th century.
@UlrikeFischer makes a change from blaming Joseph
3:09 PM
@UlrikeFischer Ah, I see! Must check, whether ocgx2 can cope with floats that sneak into open PDF Layers.
We interrupt this TeX-related discussion to bring important news from the duck world: quack!
Now, back to our normal schedule.
@AlexG btw I saw your question where you patch hyperref to inject ocg code tex.stackexchange.com/questions/477878/…. I need to put stuff in links too and one part of the pdfresources project is to add sensible hooks to links.
@UlrikeFischer I see, you are aware of this already.
@UlrikeFischer You intend to write a generic interface for starting and ending links, right?
3:32 PM
@AlexG not really, currently I'm converting the luatex driver of hyperref so that the internal definitions have sensible hooks and use l3driver commands instead of primitives. And then I try out if tagpdf can use the hooks and correctly tag the links. Joseph writes the l3driver commands and so is more in the "generic interface".
@UlrikeFischer I would like to help if need be. But I am not sure whether I really have something to add.
@JosephWright The Elements, in Haiku.
@PauloCereda What are the features that distinguish important quacks from unimportant ones? Or are all quacks equally important?
@AlexG you certainly can help, your code contains almost everything that should actually be in expl3, and you probably also know all pdfresources where packages can clash (like catalog entries, or colorspaces, or transparancy) and where some central management is needed. The idea is simply that instead of you defining a \pbs_pdfemc: and me a \__uftag_mc_emc: and accsupp a \ACCSUPP@emc for all sort of engines there should be a driver command and we all can use it.
3:48 PM
@UlrikeFischer Yes, that would be great. Something well tested at a central place.
@AlanMunn ooh I never thought of it
@AlanMunn the main feature is existence, there are no important ones.
@DavidCarlisle Yes, that was my thought too. :) #livinguptothemeanreputation
@DavidCarlisle oh no
@AlanMunn gotta love the hashtag :)
@PauloCereda unfortunately this is stackexchange not twitter so it's a mis-placed octothorp not a hashtag.
4:01 PM
@DavidCarlisle ooh
@DavidCarlisle I read on Twitter that it is good to be friend with @PauloCereda
@koronkebitch 1. Don't read the TeX book, read sth specifically for LaTeX 2. Get into good habits early on (eg, don't use $ for maths) 3. Use AMSLaTeX 4. Visit http://tex.stackexchange.com, the people there are super-helpful and some of the key LaTeX ppl are there 5. @paulocereda is great to know
@mickep oh
4:18 PM
@AlexG if you have suggestions or code or ideas or pull requests or whatever make them either here to me and @JosephWright or at the pdfresources github or at the latex3 github or per mail ...
@mickep I am but a humble idiot on the internet. :) I verbosely talk, make a fool of myself all the time and probably suffer from Pollyanna syndrome. Besides, I like to anthropomorphize onomatopoeically duck voice descriptions. :)
@PauloCereda @AlanMunn @DavidCarlisle thespruce.com/sounds-ducks-make-386223
@PauloCereda Sounds sympathetic to me :)
@DavidCarlisle /honk
@mickep never believe what you read on the on the internet
4:28 PM
@mickep -- But (perhaps sadly) there is now, strictly speaking, no AMSLaTeX. This has officially been subdivided into amsmath and `amscls'. Otherwise mostly good advice.
@PauloCereda dinner!
@DavidCarlisle oh no
But it worked. :)
@barbarabeeton Oh, I see. I never understood what is what when it came to ams*.
@barbarabeeton While you are here. I noticed in the style guide that it suggests to set Theorems unindented, with the word "Theorem" (and its number) in bold. If I do not remember things wrongly, it used to be indented (and typically set in small caps instead of bold). In AMS journals like Trans. and Proc. ..., that is. Do you happen to know when and why that changed?
4:51 PM
@mickep -- You're correct that the style changed from caps/small caps (indented) to bold (flush left). Random checking of posted articles finds that the change took place in the late 1980s (small caps in the last 1987 issue of Proceedings, bold in the last 1988 issue). The current AMS style guide specifies bold., as you point out. I don't know the reason, but (from my own curiosity), I'll ask.
5:34 PM
@DavidCarlisle ^^ We got theatre tickets!
@UlrikeFischer the play can't be any sillier than real life
@DavidCarlisle and like in real life we had to move the date to get there a few times ;-)
@UlrikeFischer if you go and find out how it ends, let Mrs May know.....
@DavidCarlisle I think she is in the play: theater-kr-mg.de/spielplan/inszenierung/…
5:53 PM
@UlrikeFischer ooh where's the Bär ticket? :)
@PauloCereda I can smuggle him in for free in my handbag.
@UlrikeFischer ooh stealth bears :)
@mickep Well, the $ business is a bit more complex
@AlexG As @UlrikeFischer says, I'm primarily on the low-level stuff, with the idea being that we (team) deal properly with as much PDF resource stuff as we can. There may be a few edge cases for things like media9, but really in the end if I have to expose an official interface for \pdfliteral, that's how it is
6:36 PM
@JosephWright did you see the scaling/accuracy question?
6:49 PM
@JosephWright I agree about that (I always write $ for inline math).
@barbarabeeton Oh, nice! I'm "staying tuned".
7:19 PM
@barbarabeeton @mickep I got curious and had a look at Proceedings, and there it seems the change happened between vol 104 (1988) and vol 105 (1989).
@HaraldHanche-Olsen -- Yup. That's what I found. I've asked one of the "old timers" if she remembers the reason for the change. Will report if she answers.
@UlrikeFischer No?
@JosephWright shocking news: ducks can also honk
@mickep The team agree too :)
@HaraldHanche-Olsen @barbarabeeton Thanks! One reason I asked is that the "old" indented in caps style is suggested by Lansburgh in his old classic book "Sättningsregler". (I'm writing a text with a colleague, and we are both fond of the formulations of typographical "rules" in that book, and we follow them where we see them (if we can))
7:22 PM
@PauloCereda /Honks in dispair
@JosephWright oh no
@UlrikeFischer OK, so next page related stuff?
@mickep -- Most of my books are still in boxes after my move. I've just checked the "updated edition" of Swanson's Math into Type, published in 1999, and online. That specifies that theorem headers are flush left and bold. That coincides temporally with the style change, and I remember that there were other changes, including the decision to apply "editorial lite", that would have been anathema to Ellen Swanson. When I unpack, I'll check the previous editions, and also Chaundy (Oxford).
@barbarabeeton Thanks! I guess the "editorial lite" was because of $$ (no, not (displayed) math mode).
@JosephWright here: tex.stackexchange.com/questions/485339/… I thought you would perhaps enjoy the wanted accuracy ;-)
@JosephWright ? what does this refer too?
7:38 PM
@UlrikeFischer BDC/EMC/... support
@JosephWright ah, not directly related to next page or not. That's simply something that is inserted very often. On every page I'm inserting some, and accsupp inserts some and ocgx too.
@UlrikeFischer Sorry, I meant 'Next on the to do list: support for page-based specials such as BDC'
@UlrikeFischer First, though, I'm thinking about quadpoints
@JosephWright oh. english slang ;-)
@JosephWright I think I can write most of the emc/bdc commands - most can be copied from pdfbase and accsupp. You then only need to check the names and look for errors and copy at the right place
@UlrikeFischer Sure, it's easy code to write: I'm trying to make sure that we have a full set of stuff ready for the next release, that's all
@UlrikeFischer I've got one oddity in the dvips link breaking to examine (fix?), plus I'm not 100% happy about the 'model box' stuff there
@UlrikeFischer I'd also like to know how widespread quadpoint support is: is it worth trying to use this for link breaking generally
@JosephWright you mean in the pdf viewers? I don't know, but it looks naturally better for larger areas (with border) than a bunch of line links.
8:01 PM
@UlrikeFischer In Adobe Reader I still get a border for each line, but I think the structure would be clearer
@JosephWright hm. I think I saw an example with pdf comments where quadpoints gave better results.
8:22 PM
@UlrikeFischer I can believe that
9:00 PM
Could you ducks please try to quack a bit less? It is still hibernation time! ;-)
1 hour later…
10:15 PM
@JosephWright @AlexG I'm not getting it how to write a BMC operator with dvips. BDC works fine - both with the pdfbase and the accsupp syntax, but the BMC is always missing in the pdf (and preflight complains ...):
\special{ps:[/Artifact /BMC pdfmark}
some text
\special{ps:[/EMC pdfmark}

\special{ps: mark /Artifact /BMC pdfmark}
some text
\special{ps: mark /EMC pdfmark}

\special{ps:[/Artifact <</MCID 0>> /BDC pdfmark}
some text
\special{ps:[/EMC pdfmark}

\special{ps: mark /Artifact <</MCID 0>> /BDC pdfmark}
some text
\special{ps: mark /EMC pdfmark}

@UlrikeFischer Probably I should also look at catalogue entries ...
@JosephWright Could you help me build PGF using l3build? (not right now, of course)
10:31 PM
@HenriMenke Sure
@HenriMenke I'll have to look at the structure: probably you are going to need some new features (for example, probably some kind of flexible file map)
@JosephWright I browsed the l3build manual, but was kind of confused because all of it assumes that one uses docstrip.
@JosephWright The PGF repo is already in TDS format, so I had hoped that it would be easier to build it.
@HenriMenke Works without DocStrip too: for example used for beamer
@HenriMenke Ah, right: that should be similar to beamer then

« first day (3090 days earlier)      last day (689 days later) »